Organ preservation after neoadjuvant immunotherapy in mismatch repair-deficient colorectal cancer: response assessment, watch-and-wait, and salvage surgery

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Neoadjuvant immune checkpoint blockade has produced high pathological and clinical response rates in mismatch repair-deficient (dMMR) or microsatellite instability-high (MSI-H) colorectal cancer, raising the possibility of response-adapted organ preservation. However, the evidence is anatomically uneven and remains immature for routine surgery avoidance. In dMMR colon cancer, neoadjuvant studies primarily establish drug activity through resection-confirmed pathological and nodal response; reliable criteria for nonoperative assessment have not been validated. In dMMR rectal cancer, clinical complete response (cCR) after PD-1 blockade may support watch-and-wait in highly selected patients because local response can be reassessed using digital rectal examination, endoscopy, pelvic MRI, biopsy when appropriate, systemic imaging, and tumor markers. This narrative review critically appraises the biological rationale, prospective and observational evidence, response-assessment limitations, surgical decision points, surveillance requirements, toxicity, and unresolved controversies surrounding organ preservation after neoadjuvant immunotherapy. We use the term immune-response-adapted surgical deferral to emphasize that surgery remains available for incomplete response, unresolved uncertainty, or regrowth. Current implementation should be confined to experienced multidisciplinary programs with rigorous molecular confirmation, standardized multimodal assessment, reliable follow-up, informed patient participation, and timely salvage capacity. Particular attention is given to immunotherapy-specific uncertainty in defining durable cCR, the limitations of circulating tumor DNA (ctDNA) as a stand-alone local decision tool, hereditary-risk considerations in Lynch syndrome, and the need for location-specific prospective validation.